A new framing for Behavioural Science
We welcome the call for ‘s-frame’ thinking and reflect on ways this offers a positive direction for the discipline
A recent paper by Nick Chater and George Lowenstein has argued that behavioural science placed too much emphasis on changing individual behaviours (with "disappointingly modest" results) and has not placed enough focus on wider behaviour change through systemic measures such as regulation and taxation. They characterise these, respectively, as the ‘i’-frame, (interventions aimed at the individual’s behaviour), and the ‘s’-frame (systemic interventions like rules, laws and taxes).
While this has resulted in plenty of discussion, others might reasonably wonder what took so long to come to this conclusion. After all, those that are used to behaviour change systems such as COM-B or MAPPS, have long understood the need to focus on both individual and wider environmental factors. Certainly, both of these are typically considered in most programmes of work. Within COM-B it is the ‘Opportunity’ element, within MAPPS it is ‘Physical’ but also ‘Social’ dimensions. At the stage of ‘Diagnose’, (when identifying and measuring the mechanisms that are shaping behaviour), then many applied behavioural science practitioners are very familiar with the need to look at the ‘s-frame’.
The challenge perhaps comes at the ‘Design’ stage, when considering the different tools (or interventions) that might be available to influence behaviour. It is tempting to assume these are limited, with Chater and Lowenstein suggesting interventions such as taxation and regulation. For the commercial sector the equivalent may be seen as, for example, distribution strategy. Hardly a surprise therefore that behavioural scientists have shied away from this. These areas are often not in their purview but also stakeholders engaging behavioural scientists do not always have the power to deliver on these changes.
However, we consider this view of the ‘s-frame’ as limited. Surely this does not only include ‘Physical’ concerns (e.g. regulatory, taxation, distribution) but also Social and Cultural considerations. We sit in a ‘system’ of people and their behaviours, which shapes what is acceptable or expected of us, creating social and cultural norms, beliefs, and expectations. In many ways these can be more powerful than rules and regulations – as the reality is that many of these are unenforceable if people simply choose to ignore or evade them. With this in mind, the opportunities for facilitating change through this ‘s-frame’ is much greater than Chater and Lowenstein seem to be suggesting. And not only that, but this is a space in which behavioural science has much to offer.
We have written previously about the importance of ‘transformational’ change activity where there is less of a push for concrete gains, and the focus is on altering the ‘climate of public debate’ to make much more far-reaching changes possible. This is the space in which politicians, marketers and activists operate. As an example, look at the way in which Apple advertising famously influenced the wider set of expectations of computing. Arguably this fundamentally changed the wider ‘s-frame’ environment and deeply influenced the general public’s beliefs, attitudes and behaviours towards computing devices.
As behavioural scientists we are familiar with the tools that are needed to Design effective communications, we have the understanding and the tools that allow us to operate in the ‘s-frame’ zone.
In addition, we suggest the ‘i-frame’ and ‘s-frame’ are not binary (as might be inferred from a reading of the paper). We can find ways to influence the ‘s-frame’ by working on changes in the ‘i-frame’ with individually oriented interventions which then have wider spill-over and catalyst effects in the wider population.
Implications for testing
This discussion does of course have implications for how we go about testing the effectiveness of interventions. The RCT is often considered the ‘gold standard’ of testing, following the biomedical model of assessing the impact of different treatments on representative samples of the population. RCTs can be very useful when assessing the potential effectiveness of an intervention for which a very tangible, bounded decision is sought. In other words, they work well for ‘i-frame’ style problems, as Chater and Lowenstein point out.
But of course, this approach to testing policy and marketing strategy initiatives fails to recognise the ‘s-frame’, the complex way in which behaviours are inter-related, sitting in a broad eco- system of influences and considerations. An RCT typically isolates one element of this eco-system, meaning that the design (and subsequent evaluation) of the intervention is limited. These themes are well-known to those engaged in changing behaviours, as evidenced by papers such as Cartwright and also by Ogilvie. Indeed, Ogilvie suggests there in an ‘evaluative bias’ in favour of behavioural interventions that are easier to test.
So how should we evaluate?
We consider that this more holistic understanding of behaviour (‘i-frame’ and ‘s-frame’) requires us to move away from an over reliance on RCTs. While they can fulfil some testing and evaluation needs, they are far from a comprehensive solution. To reflect the wider mandate of intervention activities, we suggest the following:
Using an effective ‘Diagnosis’ of the behavioural mechanisms that are shaping the behaviour, we can use a range of market and social research methods to assess the extent to which any intervention is influencing them. While this is not directly measuring a change in behaviour, we can collate evidence that these mechanisms shape outcomes and as such be confident that the interventions successfully influencing these are likely to change the target behaviours.
There is a case for a greater focus on natural experiments – what has worked elsewhere in the past? If we can compare different sorts of intervention activities and link them to the likely behavioural dimensions that they are leveraging, then we can start to make assertions about what types of intervention are likely to be helpful moving forwards
Using ‘n-of-1’ studies (small samples in a longitudinal setting) we can more easily get an understanding of ‘what works’ but also importantly why this works.
The wider theme here is that of setting aside a very narrow and a-theoretical measure of success as measured by an RCT and instead embrace the way in which a solid and well-informed understanding of the behaviour, linked through to intervention design in a coherent way, is able to offer a much more flexible and, in all likelihood, valid means of evaluation and testing.
In conclusion
The ‘i-frame’ often involves a very stripped back, individualistic notion of human behaviour, characterised as operating outside of a range of environmental influences. This has never been all that convincing for many. The ‘i-frame’ model all too often takes us down the route of limiting ourselves to cognitive processes and experimental designs (RCTs) that limit the influence of the ‘s-frame’ when we test the impact of interventions on behaviours. Of course, in some contexts this is appropriate and helpful but as the issues that we seek to influence become ever more complex, then this is ever harder to justify.
A more holistic approach requires not only draws on a wider body of knowledge (ranging from social cognition to socio-cultural psychology) encompassing both the ‘i-frame’ and the ‘s-frame’. But also we can see the way in which social and market research methods become key tools for behavioural scientists. Change is mediated through the wider social and cultural climate, and as such marketers and communications specialists are often key collaborators. Success is measured by surveying the public’s attitudes, beliefs, goals and values, just as much as through immediate changes in behaviour.
We could consider the Chater and Lowestein paper an ‘s-frame’ intervention as it certainly seems to be altering the ‘climate of public debate’ concerning the way behavioural science is understood. Indeed, as they themselves say:
“… if the right s-frame solutions were available but not implemented all along, it is likely that behavioral scientists’ enthusiasm for the i-frame has actively reduced attention to, and support for, systemic reform, as corporations interested in blocking change intend. We have been unwitting accomplices to forces opposed to helping create a better society.”
This certainly seems to be a rallying cry for a reframing of behavioural science.